Nonlinear system identification using modified variational autoencoders
This research proposes a methodology for identifying nonlinear systems using input/output data and deep learning generative models. Our framework integrates Variational Autoencoders (VAE) with Nonlinear Autoregressive with exogenous input (NARX) in a unified identification structure to address overf...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2024-06-01
|
| Series: | Intelligent Systems with Applications |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2667305324000206 |
| _version_ | 1797289666477555712 |
|---|---|
| author | Jose L. Paniagua Jesús A. López |
| author_facet | Jose L. Paniagua Jesús A. López |
| author_sort | Jose L. Paniagua |
| collection | DOAJ |
| description | This research proposes a methodology for identifying nonlinear systems using input/output data and deep learning generative models. Our framework integrates Variational Autoencoders (VAE) with Nonlinear Autoregressive with exogenous input (NARX) in a unified identification structure to address overfitting in nonlinear system identification using NARX structures. Specifically, we modify a variational autoencoder by replacing the decoder module with a NARX model using the latent space information captured from the VAE encoder module as one of the exogenous inputs. Following the training phase, the decoder module can be used as a nonlinear model of the system. We evaluate the efficacy of our approach by performing open-loop prediction tests on data from four nonlinear benchmark systems: Cascaded tanks, Gas furnace, Silverbox, and Wiener-Hammerstein. The proposed VAE-NARX method reported Root Mean Squared Error (RMSE) of 8.23×10−3, 16.69×10−3, 0.002×10−3 and 0.037×10−3 respectively. Our results demonstrate that our proposed method achieves similar and outperforms prediction performances to standard identification techniques and can enhance the performance of traditional nonlinear system identification methods based on multi-layer perceptron models. |
| first_indexed | 2024-03-07T19:08:24Z |
| format | Article |
| id | doaj.art-76c1922c1b7346348a42286a844214d3 |
| institution | Directory Open Access Journal |
| issn | 2667-3053 |
| language | English |
| last_indexed | 2024-03-07T19:08:24Z |
| publishDate | 2024-06-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Intelligent Systems with Applications |
| spelling | doaj.art-76c1922c1b7346348a42286a844214d32024-03-01T05:07:42ZengElsevierIntelligent Systems with Applications2667-30532024-06-0122200344Nonlinear system identification using modified variational autoencodersJose L. Paniagua0Jesús A. López1Corresponding author.; Facultad de Ingeniería, Universidad Autónoma de Occidente, Cali 760030, ColombiaFacultad de Ingeniería, Universidad Autónoma de Occidente, Cali 760030, ColombiaThis research proposes a methodology for identifying nonlinear systems using input/output data and deep learning generative models. Our framework integrates Variational Autoencoders (VAE) with Nonlinear Autoregressive with exogenous input (NARX) in a unified identification structure to address overfitting in nonlinear system identification using NARX structures. Specifically, we modify a variational autoencoder by replacing the decoder module with a NARX model using the latent space information captured from the VAE encoder module as one of the exogenous inputs. Following the training phase, the decoder module can be used as a nonlinear model of the system. We evaluate the efficacy of our approach by performing open-loop prediction tests on data from four nonlinear benchmark systems: Cascaded tanks, Gas furnace, Silverbox, and Wiener-Hammerstein. The proposed VAE-NARX method reported Root Mean Squared Error (RMSE) of 8.23×10−3, 16.69×10−3, 0.002×10−3 and 0.037×10−3 respectively. Our results demonstrate that our proposed method achieves similar and outperforms prediction performances to standard identification techniques and can enhance the performance of traditional nonlinear system identification methods based on multi-layer perceptron models.http://www.sciencedirect.com/science/article/pii/S2667305324000206System identificationDeep learningGenerative modelingNonlinear dynamic systems |
| spellingShingle | Jose L. Paniagua Jesús A. López Nonlinear system identification using modified variational autoencoders Intelligent Systems with Applications System identification Deep learning Generative modeling Nonlinear dynamic systems |
| title | Nonlinear system identification using modified variational autoencoders |
| title_full | Nonlinear system identification using modified variational autoencoders |
| title_fullStr | Nonlinear system identification using modified variational autoencoders |
| title_full_unstemmed | Nonlinear system identification using modified variational autoencoders |
| title_short | Nonlinear system identification using modified variational autoencoders |
| title_sort | nonlinear system identification using modified variational autoencoders |
| topic | System identification Deep learning Generative modeling Nonlinear dynamic systems |
| url | http://www.sciencedirect.com/science/article/pii/S2667305324000206 |
| work_keys_str_mv | AT joselpaniagua nonlinearsystemidentificationusingmodifiedvariationalautoencoders AT jesusalopez nonlinearsystemidentificationusingmodifiedvariationalautoencoders |